Ferrous metals are the basic material which use in various industries: in mechanical engineering, aircraft construction and shipbuilding, for manufacturing designs from metal for industrial and housing construction, transport communications and technics. At an output on the international market reception of properties of this or that metal becomes even more actual as allows to find an equivalent domestic substitute to foreign marks of steel both on structure, and on properties.

In Donbass there is three most known and leading metallurgical factories a of Donetsk: DonetskSteel, DMF and ISTIL. - modern metallurgical actions which in a year let out nearby 1 million tons of liquid steel and give All of them production not only in the countries of the former UIS, but also abroad. Thus production should be qualitative to not lose clients and to keep the competitiveness in the world market. And for this purpose one of the prominent aspects is the control over quality of let out production.
However on all three factories there is only one central laboratory which carries out the control of conformity of the sucelted steel to the order of the buyer. Besides the control is spent on the limited number of characteristics by visual comparison of the checked sample with the standard. Therefore the problem of the analysis of images steels and alloys is considered rather actual for today. Necessity of a computerization of the given process is considered justified as it will raise accuracy of the spent analysis, will reduce time for its performance and will facilitate work of workers of laboratory.

The purpose of work is consideration theoretical bases of carrying out of the analysis of images of metal, the characteristic of the basic concepts, principles and the methods of the analysis necessary for creation of a subsystem of the analysis of images of metal. Also the purpose is the substantiation of necessity of development of the given subsystem, despite of existence of software packages which solve a problem of recognition of such images.

For construction of system of the analysis of images of cuts of steel and alloys it is necessary to solve the following primary goals:
- to analyse a production cycle of melt of steel and to analyse process of quality assurance of let out metallurgical production;
- to allocate the basic analysis stages of images lay also alloys and to analyse basic of them - a stage of definition of the size of grain of steel;
- to define the optimal methods of recognition of a structure of the image of a cut;
- to choose the optimal methods of definition of the size of grain of steel.

It is possible to consider as scientific novelty of this work application of non-standard methods for the decision of a task in view. As it it is possible to consider application Markov fields and hidden Markov models for recognition of a structure of images of cuts of various alloys. In the majority these methods are applied to recognition of persons, speech and the facsimile text. Also it is possible to consider as novelty realization of process of definition of the size of grain by four ways, instead of to two as it is realized in existing systems.

The basic investigation phase of photos of metal is the schedule of metals - allows to receive the information on morphology of elements of structure. Stages schedules of metals:
1. Electronic microscopy. Usually scanning electronic microscope in which the bunch ýëåêòðîíîâ by means of system of development scans a surface of the sample is applied to research of metals.
2. Micromechanical tests. The basic characteristic of metals and alloys are their mechanical properties, ò. å. Ability to maintain loading and to be deformed. A plenty of standard methods of mechanical tests is developed.
3. Revealing and definition of size of grain of an alloy.

Grains of metals are the separate crystals of a polycrystalline conglomerate divided among themselves by adjacent surfaces, named borders of grains. At presence of doubles by grains consider crystals together with doubles. The size of grain is an average size of casual sections of grains in a plane øëèôà metal. The photo of a cut of an alloy is presented below on fig. 1.

Figure 1 Cut of steel.

The size of grain is defined by means of the methods described in the metallurgical standard.

Structures are the important characteristic of images of natural objects and the important element of visual perception. The significant amount of researches which essential part is borrowed with a problem of segmentation of images to textural attributes is devoted to the machine analysis of textural images. In this case for recognition of structures will be used Ìàrkov casual fields and hidden Ìàrkov fields.

Ìàrkov models are powerful means of modelling of various processes and recognition of images. By nature Ìàrkov of model allow to consider existential characteristics of signals and consequently have received a wide circulation in recognition of speech, and recently and images (in particular persons).

Each model represents set N of conditions , between which transitions are possible. In each moment of time the system is in strictly certain condition. In the most widespread Ìàðêîâñêèõ models of the first order it is necessary, that the following condition depends only on a current condition.

In recognition of images hidden Ìàðêîâñêèå models are applied as follows. To each class i there corresponds the model . The distinguished image is represented in the form of sequence of supervision Î. Then for each model the probability of is calculated that this sequence could be generated by this model. The model , which has received the greatest probability, is considered the most suitable, and an image carry to a class j.

The most important problem in the analysis and classification of the image for today is classification of structures which are located under various corners and in different scale. This problem can be solved by means of versions Markov models. Application of Ìàrkov models has a number of advantages:
1. The mathematical structure hidden Ìàrkov models is very rich and allows to solve mathematical problems of various areas of a science.
2. Competently designed the Ìàrkov model yields good results of work in practice.

Basic advantage of Ìàrkov models consists that the marking and each structure is characterized by the Ìàrkov casual fields, not dependent from each other. It allows to modify distinguishing system easily. Introduction of a new structure or removal existing occurs within the limits of model very simply. Besides in the given model aprioristic knowledge of a marking are easily considered.

1. Sorokin V. G., Steel and alloys: the marker / V. G. Sorokin, M. A. Gervasev's - Moscow 2001;
2. Solntsev U. P. Steel and alloys: the directory/. U. P. Solntsev - 2003;
3. Resources of Internet:
- www.ngmk.info;
- www.yandex.ru;
- www.vikipedia.ru;
4. Tsopkalo N. N. Mathematical programming and recognition of images: article / N.N.Tsopkalo - 2006;
5. Yakovlev À. V. Methods of the analysis and synthesis of structures: article / À. V. Yakovlev – 2006;
6.Goncharov À. V. Development and research of algorithms and methods detection and identification of images: article / À. V. Goncharov - 2006.

home